PG-Git (Semantic Memory MCP)

A persistent, PostgreSQL-backed repository management system and semantic memory MCP server. Instead of storing Git objects loosely on the file system, PG-Git stores the entire Directed Acyclic Graph (DAG) natively in PostgreSQL, complete with automatically generated, temporally-decayed semantic vector embeddings for AI IDEs.

🧠 Why PG-Git?

In the standard AI coding agent ecosystem, searching codebases relies on rigid grep searches or expensive AST parsing. PG-Git fundamentally changes this by bridging Git directly with Vector Databases:

1. Semantic Code Search: Find code based on what it does, not just its syntax.

2. Exponential Temporal Decay: PG-Git mathematically decays older vectors. Your agent will prioritize code you wrote yesterday over highly similar dead code written 6 months ago.

3. Local-First Purity: No cloud APIs. It uses Ollama with nomic-embed-text for 100% private, on-device vectorization.

4. ACID Compliant: Native transactions ensure complete safety for multi-node accessibility and concurrent AI swarm agents.

⚡ Quick Start

You must have Ollama running with the nomic-embed-text model pulled:

ollama run nomic-embed-text

1. Install Dependencies & Migrate You will need a running PostgreSQL instance with pgvector enabled.

npm install
cp .env.example .env
# Edit .env with your PostgreSQL credentials
node db/migrate.js

2. Import Your GitHub History You can instantly import any local .git repository. PG-Git will natively parse the Git history, generate semantic embeddings for all blobs, and securely deduplicate them into PostgreSQL:

npm run import

3. Add to your Agent / IDE Configuration (e.g. mcp_config.json):

{
  "mcpServers": {
    "pg-git-mcp": {
      "command": "node",
      "args": ["/absolute/path/to/pg-git/server/mcp.js"],
      "env": {
        "OLLAMA_URL": "http://localhost:11434",
        "EMBED_MODEL": "nomic-embed-text"
      }
    }
  }
}

4. Start the Web UI (Optional) PG-Git includes a sleek, dual-pane IDE interface for browsing your semantic repositories.

npm run dev

🚀 Real-World Usage Examples

To effectively use PG-Git, simply speak to your IDE agent normally. It will use the MCP tools (pg_git_semantic_search, pg_git_list_repos, pg_git_read_tree, pg_git_read_blob) to interface with the database.

Example 1: Finding specific logic

You: "Where do we handle the temporal decay for the memory MCP?" Agent: [Calls pg_git_semantic_search] "I found the logic in server/index.js inside the krusch-memory-mcp repo. It uses the exp(-DECAY_RATE * age_in_days) formula."

Example 2: Reading a repository tree

You: "What is the folder structure for the pg-git project?" Agent: [Calls pg_git_read_tree] "Here is the root directory structure..."

How Does Temporal Decay Work?

When calling pg_git_semantic_search, PG-Git returns the highest cosine-similarity matches. However, it applies Exponential Temporal Decay based on the blob's last_seen_at timestamp. If you have two very similar pieces of code, the newer one will have a significantly higher score, preventing your agent from hallucinating based on outdated implementations.

🤖 The Autonomous Agent Workflow (/close)

You can integrate PG-Git into your agentic workflow to ensure your semantic memory is always up to date.

Whenever you step away from a task, tell your agent to run the snapshot script. The agent will autonomously:

1. Hash the current project folder into Git Blobs and Trees.

2. Ping Ollama to embed any new or modified files.

3. Commit the state directly into PostgreSQL.

Command to run:

npm run snapshot

🛠️ Configuration & Environment Variables

License

MIT License. Created by kruschdev.